1: \begin{abstract}

2: We investigate the effect of numerical magnetic diffusion in magnetohydrodynamic simulations of molecular clouds. 

3: To do this we performed convergence studies on adaptive-mesh isothermal simulations of sub-critical molecular clouds.

4: We found that the simulations with low and middle resolutions collapse, contrary to what is expected, however the simulation with

5: the highest numerical resolution ossilates around an equilibrium state without collapsing. The requirement for avoiding

6: the spurious collapse was to solve the Jeans' length with at least 8 cells.

7: 

8: %These results confirms that the collapse of the low and medium resolution

9: %simulation were due to numerical diffusion of the magnetic field.

10: In order to quantify the numerical diffuion of the magnetic field, we ran a second suit a simulations aimed at 

11: measuring the numerical magnetic diffusion coefficient. Afterwards, we computed the corresponding diffusion times

12: for a given numerical resolution. This results let us propose a characteristic-time-based resolution criterion

13: for the correct resolution of magnetic fields. This criterion results from requiring that the diffusion time is 

14: larger than the relevant dynamical time to the physical process responsible for the dynamic evolution of the structure; for sub-critical clouds,

15: the free fall time.

16: 

17: %This is a simple template for authors to write new MNRAS papers.

18: %The abstract should briefly describe the aims, methods, and main results of the paper.

19: %It should be a single paragraph not more than 250 words (200 words for Letters).

20: %No references should appear in the abstract.

21: \end{abstract}

22: